1. Technical Field
The present invention relates to a teleconference terminal apparatus, a relaying apparatus, and a teleconferencing system performing speech communications between plural terminals via an IP network.
2. Background Art
Recent development of an IP network has made possible image and speech communications on the IP network. This has commercialized a multipoint teleconferencing system to realize image and speech communications between multi points.
The techniques connecting between each of terminals in the multipoint teleconferencing system are typically classified into the mesh network and the star network. FIG. 1 shows the connecting techniques.
It is noted that the drawing illustrating the mesh network in FIG. 1 shows the point A-centered connection relation alone for simplicity, and a connection relation with another terminal centered is omitted.
In the case where there are n terminals, the mesh type allows each of the terminals to establish a direct connection to another N−1 terminal without routing through a relay point. In this case, a stream traveling through the network does not route through an unnecessary relay point. This makes possible avoiding a further stream delay however more the number of terminals increases.
In the mesh type, however, each terminal needs to decode the stream from the N−1 terminal. Hence, the number of connection points is required to be limited depending on terminal performance. Further, the mesh type presents a drawback of causing packet delay due to increasing network traffic.
In the star type, streams transmitted from each of the terminals are once brought together to an MCU (Multipoint Control Unit). Then, the MCU decodes and mixes the streams, and the mixed speech signal is coded and transmitted to each of the terminals.
In this case, since decoding a stream of the mixed speech signal, each terminal is free from a load however more the number of terminals increases. Routing through the MCU, however, develops a stream delay. Moreover, the star type presents a drawback of overloading the MCU.
As described above, the multipoint teleconferencing system, employing either the mesh-type connecting technique or the star-type connecting technique, tends to significantly overload each of the terminals or the MCU. Thus, another technique has been proposed to delete speech data based on information of speech amplitude in order to alleviate network overload and processing amount in decoding (See Patent Reference 1, for example).    [Patent Reference 1: Japanese Unexamined Patent Application Publication No. 2000-252981]
As the multipoint teleconferencing system has become commercially available, users have started to request more participants to a teleconference with the use of more terminals and higher sound quality which was unavailable before.
In particular, realizing a teleconferencing system having high sound quality needs to employ the MPEG (Moving Picture Experts Group) Audio used for music, rather than a speech codec having a narrow bandwidth.
Deleting an unnecessary stream as shown in the Patent Reference 1 in the conventional technique requires a speech codec, such as G. 729 including information on speech amplitude.
It is noted that a stream in the MPEG Audio also includes gain information; that is, a kind of information on speech amplitude. Accurate speech amplitude, however, is not available unless the stream is decoded once.
In other words, obtaining the accurate speech amplitude out of a stream which conforms to the MPEG Audio standard causes further processing overloading; that is, decoding the stream for the accurate amplitude.
In addition, even though the data amount of streams transmitted from each of the terminals reduces, each terminal needs to decode in parallel as many streams as the other terminals. Thus, increasing number of terminals directly results in further processing overloading on each of the terminals.